modernize-harden
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes source code from legacy systems, creating a surface where malicious instructions embedded in the code could attempt to influence the agent's behavior during the scan. It includes a specific defense mechanism to identify 'instruction-shaped text' (injectionFlags) to mitigate this risk.
- Ingestion points: Source code read from legacy/$system.
- Boundary markers: Lacks explicit separators for input code, but utilizes injectionFlags detection.
- Capability inventory: Writes to the file system (analysis/), executes Python reporting scripts, and invokes sub-agents.
- Sanitization: Implements masking for detected credentials to prevent exposure in generated artifacts.
- [COMMAND_EXECUTION]: The skill executes system commands including readlink to resolve symlinks and python3 to run a reporting script located at a path defined by the ANTIGRAVITY_PLUGIN_ROOT environment variable.
- [DATA_EXFILTRATION]: Although the skill actively discovers hardcoded credentials, it implements a security protocol to quarantine raw values in gitignored files (SECRETS.local.md) and masks values in all shared reports to prevent accidental data exposure.
Audit Metadata